FR2634824A1 - METHOD FOR REGULATING THE WEALTH OF AN AIR-FUEL SUPPLY MIXTURE OF AN INTERNAL COMBUSTION ENGINE OF CLOSED LOOP TYPE, WITHOUT EXPLOITATION OF PHYSICAL MEASUREMENT OF WEALTH - Google Patents

Abstract

The method consists in determining an injection time ti for controlling the opening of a fuel injector, by adding to a static component tis of this time, a sampled dynamic correction DELTAtidn whose measurement is given by a linear combination d 'at least one of the DELTAtidn - k values of the dynamic corrections made at the immediately preceding sampling instants and of the estimated richness differences in - k with respect to a setpoint richness Rc, measured at the current instant and at least one instant immediately prior sampling. Application to the regulation of richness of an air-fuel mixture in the event of failure of an oxygen sensor placed in the engine exhaust gases.

Description

The present invention relates to a method for regulating the

  the richness of an air-fuel feed mixture of an internal combustion engine and, more particularly, to such a "closed-loop" type process, without the use of any physical measure of richness, designed to replace a closed loop control method of such a motor from a signal provided by an oxygen sensor placed in the exhaust gas of this engine, following a failure

  temporary or definitive of this probe.

  Processes for regulating the richness of an air-fuel mixture by modulation of a fuel injector opening time (or duration) using a multiplicative term are known, this regulation being adjusted so as to create a regular and continuous oscillation of the air / fuel ratio around a nominal value. Whether or not the engine exhaust passes into a catalytic converter that limits environmental pollution by

  these exhaust gases, the modulation of the air-

  fuel is optimized, possibly with respect to the characteristics of this pot, by means of two types of correction, proportional and integral, introduced in the calculation of the opening time of the injectors

equipping the motor to regulate.

  These corrections are generally calculated at the top dead center of the cylinder concerned of the engine. As a result, the frequency of oscillation of the air-fuel ratio is not properly controlled because it depends on the engine speed. Similarly, the amplitude of these oscillations can not be properly controlled because it depends on the magnitude of the corrections calculated and applied and, since the corrections are systematically calculated at the top dead center, it is not possible to take account of the dynamics of the system, which causes a loss in the information to be processed and therefore a lower precision of the calculations used for the control of the amplitude of the oscillations of the richness of the mixture. In the French patent application filed this day by the Applicant and entitled "Method and device for regulating the richness of an air-fuel supply mixture of an internal combustion engine", a method of regulating the closed loop, using a signal delivered by an oxygen sensor placed in the exhaust gas of the engine, which allows to completely control the frequency and amplitude of the oscillations of wealth, making them independent of the engine speed. The method implements a wealth regulation law having a recursive character, that is to say dependent on the previous states of the wealth measurements and the values of the previous corrections of the opening time.

injectors.

  Whether a proportional and integral correction method or a recurrent regulation law is used, a failure of the device for the implementation of the method is still to be feared. Very often the fault is caused by a failure of the oxygen sensor used, because this detector is placed in a particularly aggressive environment

  constituted by the exhaust gases of the engine.

  It is then necessary to provide means to ensure proper control of the richness of the air-fuel mixture, in case of failure of the wealth control normally used for this purpose and, more particularly, in case of failure of the Aa probe.

  oxygen used in this regulation.

  The present invention therefore aims to provide a

  method of regulating the richness of an air-mixture

  feed fuel for an internal combustion engine designed to operate in the absence of richness information delivered by an oxygen sensor placed in the engine exhaust. This process can then operate independently or substitute for a wealth regulation method using these wealth information, in case of temporary failure or

  definitive of this oxygen probe.

  The present invention also aims to provide such a method capable of establishing a continuous control of the richness of the mixture, from a value of reference value which may be different from that corresponding to

  stoichiometry, under certain driving conditions.

  Another object of the present invention is to provide such a method which is well protected from disturbances and allows the implementation of corrections distributed by type of corrections, so as to reduce the time required for

calculation of these corrections.

  These objects of the invention are achieved with a method of regulating the richness of an air-fuel mixture for supplying an internal combustion engine, of the closed-loop type, by controlling the opening time of an injector of fuel, this opening time consisting of at least the sum of a static injection time and a dynamic correction of the injection time, characterized in that, to form a signal representative of this correction, an estimate of the instantaneous richness of the mixture is made using a model of the response of the engine to a signal representative of this opening time, the difference of this richness is sampled at a setpoint richness, treated the sampled deviation in a sampled output corrector in synchronism with the sampling of the difference in richness to form a signal atid representative of the dynamic correction of the injection time, the sampled measurement atid, of this signal at the sampling instant n taking the form of a linear combination of at least one of the values attid._ k of the dynamic corrections performed at the immediately preceding sampling instants and the measured 6n-k estimated richness differences at the current time and at least one sampling instant immediately before. The most distant sampling time (n-k) taken into account in this linear function defines the order of the command thus obtained. According to a first implementation of the first-order method according to the invention, the measurement Atid of the sampled signal representative of the dynamic correction of the injection time is of the form:

6tid. -ttid-1 + W (eh in -

  o X and are constants adjusted according to the desired dynamics for the regulation, and e 1, the measurements of the differences of richness estimated at the

  sampling instants n and n-1.

  According to a second implementation of the method according to the invention, said second order, the measurement atidn of the sampled signal representative of the dynamic correction of the injection time is of the form: Atid. = a tidn 1 + P tidn 2 + goen + g1en- 1 + g2 in 2 oa, P, go, g1, g2 are constants adjusted according to the desired dynamics for regulation, in, e.1, e. -2 are measures of wealth disparities

  estimated at sampling instants n, n-1, and n-2.

  The static injection time tisn to which is added the dynamic correction Atidn introduces meanwhile an estimated value of the steady state injection control with corrections due to certain engine operating conditions: correction of idling, of re-coupling of 'an air compressor or other device

auxiliary, altitude, etc ...

  Other corrections are applied to the setpoint of the mixture richness to take into account other operating conditions of the engine: operation at full load, cold, acceleration

or in deceleration, etc.

  Corrections can also be applied to the static gain of the engine response model to account for certain parameters of engine operation, speed, intake pressure, air temperature,

coolant, etc ...

  Other features of the present invention

  will appear on reading the following description

  and examining the single figure which represents a functional diagram of the control method according to the invention. By hypothesis, the process according to the invention can not use a signal representative of the real richness of the air-fuel mixture such as that delivered by an oxygen probe, commonly called a "probe", placed

  in the engine exhaust.

  According to the present invention, the measure of richness that can be obtained from the signal delivered by an oxygen probe is replaced by estimating this richness, and substituting this estimate for the measurement of richness normally provided by the probe in a closed-loop regulating process, so as to retain the advantages of such stability and dynamic operation. This is in fact a Npseudo-regulation "in closed loop since the measurement of wealth used is not a physical measurement but only an estimate calculated from a modeling of the process, as it is

  will describe in more detail later.

  Thus, with reference to the single figure of the appended drawing, is it used in the present invention a model [reference representative of the dynamic relationship between the richness of the mixture and a time (or duration) of opening a fuel injector used to form the air-fuel mixture burned in an engine M & internal combustion, mixture whose wealth

  is regulated by the process according to the invention.

  If we denote by: Rc, a fixed setpoint richness for the air-fuel mixture, R, the instantaneous richness of the mixture, estimated using the model, it appears on the functional diagram of the figure that the difference of estimated richness: e = Rc-R is sampled in E1 with a constant sampling period Te, processed in a corrector C which delivers a signal representative of the dynamic correction of the injection time 6 itself added to a time of 'static' injection estimated tis, the sum ti being sampled in E2, in synchronism with the sampling El. The signal ti is representative of an effective resultant injection time, used with the model for the establishment of the wealth According to a preferred embodiment of the present invention, the corrector C used is a digital corrector. Motor modeling requires the determination of a dynamic modelM (p) such that: _K (p) = Gs.Md (p) (1) where Gs is the static gain of the steady-state model, Md (p) characterizes the dynamic of the response of the richness R according to the control signal ti,

p, the Laplace operator.

  In steady state we have: R = Gs.ti with: ti K.R.R, where K is a characteristic coefficient of the engine and operating conditions, R is the desired richness of the mixture, set at R = 1

RF the fill factor.

  It is shown that: Gs = 1 / K (N, Pr) .g0 (Pr) where N is the speed, or speed, of the engine, Pr the pressure at the intake manifold of the engine, go = a. [Pr + p0],

ao and po constants.

  Measurements made on the engine on the bench make it possible to establish a map in the pressure system Pr / speed N, values of the coefficient K. The sampling of the expression (1) above of the dynamic model gives: Z [BoM (p) J - Gs.Md (Z) where Md (Z) is the Z transform of the dynamic model, Bo the "zero order blocker" function introduced by

sampling.

  The Md (Z) model is then a recurrent model that makes it possible to estimate the instantaneous richness at the sampling instant n as a function of the actual injection time ti .: Rn = Gs (N, Pr). Md (Z) .ti, which gives the sampled richness gap: e. With the method according to the invention, it is possible to give any value chosen to the setpoint richness. For example, it will be possible to fix: Rc-i for a depollution operation, Rc> 1 during an acceleration period, Rc <1 for an operation with a mixture

  "poor", also for the purpose of depollution.

  If a catalytic converter is associated with the engine to treat the exhaust gas, it is known that a good operation of this catalytic converter requires an oscillation of the richness of the mixture. For this purpose, according to the invention, a periodic variation as a function of time of the setpoint richness Rc is programmed in the form of a square signal, for example, an alternating signal.

  on both sides of the desired average value.

  If we denote by C (Z) the Z-transform of the transfer function of the corrector C used in the present invention, we have:

t t '- C (Z) .e.

  and the actual injection time, sampled, is given by: A

tin tis. + Atid.

  o tisn is the estimated static injection time which introduces possible corrections such as: - an idle speed correction, - a coupling correction to the engine of an air compressor forming part of a powered air conditioning device in mechanical energy by the engine,

- a correction of altidude.

  Of course, the closed-loop dynamics will be characterized by the characteristic equation of the system: i + C (Z) .Gs.Md (Z) O- 0 Bank measurements of the richness of the mixture as a function of the injection time identify the

dynamic behavior of the engine.

  If these measurements show a dynamic relation of the first order between the richness and the mixture, one chooses for the corrector C and the model M functions

first order transfer.

  Thus, by choosing: (p) - = Gs / (1 + T p) where I is a time constant, it comes: Md (Z) - (1-a) / (Za) with a = e-T ' It is shown that such a choice provides the following recurrence laws: R aRn1 + Gs (1-a) ti, atid. = 6tid .. + '(en-in), withen = Rc- R. o Y, 7 are functions of a, of Gs and of dynamics

desired for regulation.

  If the bench measurements show a second order dynamic behavior, we choose 2nd order transfer functions for the corrector C and the model, ie for Md (Z): Md (Z) = (a'Z + P ' ) / (Z2 - 'Z + 68') which gives for the estimated wealth and the time

  of dynamic injection the recurrent relationships

  vantes: R. '' .R.1 - 6'R..2 + Gs (a'ti..l1 + P'ti._) atidn = attida. + Pttidn.2 + g0e, + gle.-. + g2eh-2, with: in Rc - R. oa, P, go, g ,, g2 are functions of a ', À', '6' and Gs (N, Pr), and of the desired dynamics in closed loop. In order to integrate the physical disturbances acting on the real engine system and affecting the static and dynamic behavior of the latter, and in order to ensure a zero steady-state static error, an integral term must be maintained 1 / (Z -1 ) in the writing of C (Z) which implies: a + 1 The signals tis. and so obtained, which constitute essential components of the total injection control time Ti of the injector must be combined with various corrections to take account of particular operating conditions of the engine or

  even of the aging of it.

  In particular the static gain Gs - 1 / K.g0 is subject to a map correction of the factor K as a function of N and Pr, as we have seen above. This correction is of the "fast" type, that is to say, capable of changing with each calculation cycle, ie at each sampling instant. An injector aging correction can also be introduced by adding a 6K term to the factor K whose estimated value then becomes: K - K (N, Pr) + 6K This aging correction is obviously

type "slow".

  Similarly, the term g 'a' (Pr + p0) can undergo a "slow" altimetry correction and a "slow" self-adaptive correction on estimated values a0 and p0 of a0 and p0. The corrected value is then used: A go ao (Pr + Po + 6p, t) Finally, air temperature corrections Cair and coolant temperature Ceau can be applied to the static gain whose estimated and corrected value then becomes : AA Gs = 1 / K.g0.Ceau.Cair Moreover various corrections can be introduced by an action on the richness setpoint value Rc, in particular: - a correction of full load, - a correction of richness at cold start - transient correction (acceleration / deceleration). A correction, in case of coupling to the engine of an air compressor for example, can also be introduced by acting on the wealth, in addition to that mentioned above introduced by acting on the duration

opening of the injector.

  All these corrections are of the "fast" type ..

  The total opening time of the injector may further comprise a term "offset" representative of a dead time of the injector control and a term 8tbat representative of a variation of the electrical voltage of the injector. supply of the injector, voltage

  provided by the vehicle battery.

  Thus, with the method according to the invention, which establishes closed-loop type control without the use of a richness signal provided by an oxygen sensor, the total duration Ti of the injector opening control signal is expressed by the relation: Ti = ti + to + tb t with: ti = tis + Atid and: Ltid = C (Z) (Rc-R) This procedure, which is implemented by means of a law recurrent control, is more particularly intended to replace the control method described in the aforementioned patent application in case of failure of the oxygen sensor used, the latter method being itself also implemented with the aid of a recurrent control law governing a sampled additive dynamic correction Ltid, which is added to an estimated static injection time to establish an injection time

resulting effective ti.

  If the failure is temporary, regulation by the method according to the present invention is abandoned upon return of the probe to its normal operation and the method of regulation of the aforementioned patent application takes over the control of the wealth regulation of the

air-fuel mixture.

  Of course, the control method according to the invention could be used independently or be associated, in the event of failure of an oxygen sensor, with a control method other than that described in the abovementioned patent application. The method according to the invention is however very complementary to the latter method in that they both implement recurrent control laws that can be executed by means; of

calculation of the same type.

  Thus, thanks to the control method according to the invention, can one avoid a complete failure of the mixture richness regulation of an internal combustion engine in the event of failure of an oxygen sensor,

  simulating the presence of this probe.

  The method according to the invention also has the advantage of allowing a continuous control of the richness, from values lower than that corresponding to the stoichiometry to values higher than the latter. The recurrent control law used is designed to maintain the dynamics and stability of the regulation despite the presence of disturbances. The corrections applied can be divided by type (slow, fast, constant) and therefore only be carried out in a timely manner from where you can see how you are doing in your own right. uosreuTqmoo aezne alno; ea Tadde zuesTe; ama: 4sAs ez; do a suep no Jnezow asseTA / sTupe aTe, p 4Tqp 8wurSAs have suep aIdumrx of snssep-To;% Toqp ana; om essse; TA / uTssTmpe, p uoTsseazd amawQsss aT enb sea = ni seu9zsXs sep suep senbFtdebo% .xeo O0 suOTZaos-Oo sep zsFTTTn ealnod uo 'npusasue ues * anaZo0w np uo4sTd a, p ineq $ = zom: uiod abessed ap uue $ su- , T u aueoiiuqo9 a DueeanbTsseTio TsToqo uo, nbsaoT seo ea sed zsa, u Tnb ao 'anazow np am g6i ap SUOzeTavA xne S eTqTsuasuT apumumOD ap TOT and aapuaa 9p amIode uo; usBAuT, I% ueAns apaood ea suep Ts-oqo e3 apo -ragd ap zue4suoo sed e sebuuoTTiugoU9, IT anb azooua ueaou uo suoT0oa = zoo s9s ap Tlnoieo p sdmwa ae suep asmouoD9 7T

Claims (16)

  1. A method for regulating the richness of an air-fuel mixture for supplying an internal combustion engine, of the closed-loop type, by controlling the opening time of a fuel injector, this opening time being constituted by at least the sum of a static injection time and a dynamic correction of the injection time, characterized in that, to form a signal representative of this correction, an estimation of the instantaneous mixing using a model of the response of the engine to a signal representative of this opening time, the gap of this richness is sampled. a setpoint richness, the sampled deviation is processed in a sampled output corrector in synchronism with the sampling of the difference in richness to form a signal representative of the dynamic correction of the injection time, the sampled measurement ttid of this signal at the sampling instant n, taking the form of a linear combination of at least one of the values tid..k of dynamic corrections performed at the immediately preceding sampling instants and of the estimated differences in richness in .k measured at the current time and at least one sampling instant immediately before. 2. Method according to claim 1, characterized in that the measurement attid of the sampled signal representative of the dynamic correction of the injection time is of the form: totid. - Atid..l + Y (in e ..) o 'and t are constants adjusted according to the desired dynamics for the regulation, en and ené, the measurements of the richness differences estimated at the moments sampling n and n-1.   3. Method according to claim 1, characterized in that the measurement & tid. of the sampled signal representative of the dynamic correction of the injection time is of the form: A A, A'-   Ltid. - attid._ 1 + PTtid._ 2 + go e. + gle- + 2e. -   o a, P, g1, g1, g2 are constants adjusted according to the desired dynamics for regulation, in, e. l, in a are the measures of wealth differences   estimated at sampling instants n, n-1, and n-2.   4. Method according to claim 3, characterized in that the coefficients a and p satisfy the relation: a + p = 5. Process consistent with any of the   preceding claims, characterized in that   said signal representative of the dynamic correction of the injection time Ltid, at a static injection time representative of the steady-state static injection control and of at least one correction of the group: idle correction, correction of the motor coupling, altitude correction, to form a signal of   control of the opening time of the injector.   6. Process according to claim 5, characterized in that a signal 8tbat is added to the signals representative of the static injection time and the dynamic correction to correct variations of the supply voltage of a controlled injector. electrical, to form a time control signal opening of the injector.   7. Process according to claim 6, characterized in that a constant offset correction (to) characteristic of the response time of the injector is added to the sum of the signals tis, Itid and ottb. control signal of the opening time of the injector. 8. Process according to any one of   Claims 1 to 7, characterized in that the model of   the response of the motor to a controlled opening time of the injector comprises a component constituted by a static gain of the form Gs l / K.go, K being a factor which undergoes a map correction in the system of admission pressure Pr / speed N of the motor,   go = a (Pr + p0) where ao and p0 are constants.   9. Process according to claim 8, characterized in that corrects the factor K by an additive correction 6K taking into account the aging of the injectors. Method according to claim 8, characterized in that an additive altimetric correction is applied 6p.I t at the end (pr + po) of g0.   11. Process according to claim 10,   characterized in that a self-correction is applied   adaptive to coefficients co and p0 of go.   12. Process according to any one of   Claims 8 to 11, characterized in that   an air temperature correction and a correction of   water temperature at static gain Gs.   13. Process according to any one of   preceding claims, characterized in that   samples the richness gap and the corrector output signal with a constant pitch Te independent of the motor rotation speed.   14. Process according to any one of   preceding claims, characterized in that it is   implemented by detecting a failure of an oxygen sensor placed in the exhaust gas of the engine and forming part of a control device in closed loop of this engine.   15. Process according to any one of   preceding claims, characterized in that   corrects the setpoint richness according to conditions   predetermined operating conditions of the engine.   16. A method according to claim 15, applied to an engine associated with a catalytic converter for the depollution of the exhaust gas of the engine, characterized in that the value of the setpoint value is regulated according to a periodic law as a function of time, alternative on both sides of the average setpoint, to ensure proper operation of the catalytic converter.